Life expectancy has nearly doubled over the past century, thanks to remarkable advances in basic science, medicine, and public health. That longer average lifespan, however, brings a greater risk of developing multiple chronic diseases at once. While so-called multimorbidity is known to be associated with poorer health outcomes-including increased hospitalizations, emergency department visits, and disability-how combinations of specific conditions and variables such as age affect people's health remains poorly understood.
At Rockefeller University, Joel Cohen's Laboratory of Populations has spent decades revealing fundamental mathematical patterns that hold true across a wide range of populations, from bacteria and insects to trees, fish, and humans. Now he and Jonathan Tobin, director of Community-Engaged Research at Rockefeller's Center for Clinical and Translational Science, and their colleagues have collaborated on a study of more than 238,000 adults to seek underlying mathematical patterns in multimorbidity that could help improve clinical practice guidelines.
They recently reported surprising findings in a paper in the Journal of Population Ageing: The older people get, the less alike their multiple chronic conditions are. This discovery raises questions about clinical recommendations based on a person's age alone. For example, the U.S. Preventive Services Task Force recommends colon cancer screening for all adults to begin at age 45 (previously age 50) and continue to age 75, while biennial screening mammography is suggested for women aged 40 to 74. Such broad recommendations may overlook complex nuances that are unique to each patient, the study finds.
"We learned that if you're writing guidelines for clinical care, it's better to base them on each individual's multimorbidity rather than on chronological age alone," says Cohen. "That's a categorical shift."
Tobin adds, "This new understanding of multiple comorbidity is critical for personalized medicine."
The "Tipping Point"
The current study grew out of Tobin's decades-long research at the intersection of medicine, science, and social factors. He is president and CEO of Clinical Directors Network (CDN), a practice-based research network that conducts clinical and health services research by engaging the clinicians and the staff of Federally Qualified Health Centers (FQHCs), which provide more than 34 million patients living in low-income communities across the U.S. with accessible, cost-effective primary care services, including medical, behavioral, dental, and mental health care, as well as vision, pharmacy, and social services. CDN's goal is to translate clinical research into clinical practice to improve health equity and public health for vulnerable communities.
Nearly a decade ago, Tobin and his collaborators began studying patients usually excluded from clinical trials and often seen at FQHCs: people with high multimorbidity. (Multimorbidity differs slightly from comorbidity in that multimorbidity tallies multiple chronic conditions, while comorbidity focuses on a primary disease that has one or more secondary, or comorbid, diseases associated with it.) The parent study is designed to test whether specific interventions could reduce such patients' high rate of hospitalization-a destabilizing and stressful experience that can set patients' health back even further, Tobin says.
But they lacked data on this cohort. "Despite being among the sickest patients, people with multimorbidity are also among the least researched," Tobin says. "They're often excluded from the studies that eventually inform clinical guidelines and clinical practice because they introduce too many confounding variables."
Tobin and his colleagues, including Mary Charlson and James Hollenberg at Weill Cornell, came up with a creative solution: Rather than exclude multimorbid patients from a study, they put them at the center of one. Focusing on people who receive services at 16 FQHCs in Chicago and New York City that are part of four health systems, they screened the de-identified health records of 238,156 people to narrow in on those with multimorbidity. Members of the research team at the FQHCs, including Community Health Network, Family Health Centers at NYU Langone in NYC, and Erie Family Health Centers and Friend Family Health Center in Chicago, reached out to patients with multimorbidity to join the Tipping Points clinical trial, which was funded by the Patient-Centered Outcomes Research Institute. In the trial, health coaches worked with the nearly 2,000 enrolled study participants and assessed whether specific interventions-especially ones that help people manage their own health before they reach the "tipping point" that sends them to the emergency room-reduced hospitalizations and disability.
Since age is a common anchor for clinical guidelines, Tobin wondered whether there were mathematical patterns underlying the progression of multimorbidity over time that might reveal when the tipping point might be reached. He asked Cohen to apply his population-mathematical expertise to the larger de-identified patient cohort of 238,156 people that was compiled from data partners including the PCORnet clinical research networks in NYC (INSIGHT) and Chicago (CAPriCORN), as well as Healthix, BronxRHIO, and AllianceChicago.
Weighting conditions
For the current study, Cohen had the age and geographic location (NYC or Chicago) of each person, as well as their score on the enhanced Charlson Comorbidity Index (eCCI), which was calculated by Hollenberg. A numerical tally designed to anticipate the risk and cost of hospitalization, the eCCI gives a person a score based on the number and severity of their chronic conditions in 39 categories. For example, myocardial infarction, congestive heart failure, and peripheral vascular disease are all rated 1, the lowest, while metastatic solid tumor, AIDS, and transplants are weighted 6, the highest. Most conditions are weighted between 1 and 3.
Cohen examined how the average and the scatter ("variance") of the eCCI scores changed with increasing age, as well as the variation of eCCI within one-year age groups, such as 70-year-olds.
Across the geographic locations, "the data show that in general, variability goes up with the average in a mathematically consistent way," says Cohen. "Specifically, the variance increases when the mean increases. That means that people's multimorbidity grows increasingly different from the multimorbidity of other people of the same age as they grow older."
This phenomenon approximates Taylor's law, which Cohen has found describes population variability in a variety of contexts, including human censuses, infectious disease, wildlife, and weather.
"In this case, for example, we found that two 70-year-old patients typically have far greater differences in their chronic conditions than two 40-year-old patients-and the variation between people will continue to grow the older the patients are," Cohen says. "That may be the reason a treatment recommended for someone based on age works for one 70-year-old but is inappropriate for another. A better course of action would be to base treatment on their multimorbidity."
"We know intuitively that as you get older, you have more chronic conditions because you have more time to accumulate them, and it's been well documented that age and multimorbidity are associated," Tobin says. "However, what wasn't examined was a mathematical model that describes well the variation in multimorbidity, and that's what Joel's analyses found."
Adds Cohen, "This new model reinforces the idea that we must embrace the complexity of each individual patient, especially when they have multiple chronic conditions. It also may help us to anticipate when someone's risk of hospitalization increases and make a specific intervention before that occurs. When it comes to medical care and public health, there's no way that one size can fit all."
